dhcpcd-discuss

Re: DHCPv6 IA_PD Renew: link-local vs Server Unicast

Roy Marples

Mon Nov 06 16:45:13 2017

On 31/10/2017 08:30, Roy Marples wrote:
It might be that the unicast reply to you is now being received for an address on a different interface. I've been working on code to handle this for a few days now, but I've just lacked the time to progress it recently.

This is already done for IPv4 due to shared IP addressing, but it looks like it's more imperitive for IPv6 because we listen to the address instead of what's actually received on the interface (our IPv6 implementation doesn't use BPF at all).

Hopefully I'll find the time to finish this soon.
In the meantime though, you could run a tcpdump on all interfaces to see if this indeed the case, or that your ISP is as you say, silent. If it *is* silent, it might be because it didn't bind the sending of the address to the dhcpv6_client port (which it cannot do becasue it's not in master mode).

This has now been implemented here:
https://roy.marples.name/git/dhcpcd.git/commit/?id=03cf61b22d7fca4f33b94ec8698f2925d7af32b3

At this point, I don't have any more todo items for releasing 7.0.
Would anyone prefer another RC at this point?

Roy

Follow-Ups:
Re: DHCPv6 IA_PD Renew: link-local vs Server UnicastNeal P. Murphy
References:
DHCPv6 IA_PD Renew: link-local vs Server UnicastJakub Jankowski
Re: DHCPv6 IA_PD Renew: link-local vs Server UnicastRoy Marples
Re: DHCPv6 IA_PD Renew: link-local vs Server UnicastJakub Jankowski
Re: DHCPv6 IA_PD Renew: link-local vs Server UnicastRoy Marples
Archive administrator: postmaster@marples.name